Date of Award

Fall 2025

Document Type

Open Access Thesis

Department

Computer Science and Engineering

First Advisor

Christian O’Reilly

Abstract

Background:Brain connectivity measures have been used to study communication between brain regions using electroencephalography (EEG). Functional and effective connectivity estimate the synchronization and the flow of information between regions, respectively. However, findings from studies using different measures to investigate similar connections do not converge. To guide the selection of functional and effective connectivity measures in future studies, we systematically compared a set of measures in the context of resting state EEG. We examined four functional connectivity metrics (coherence (Coh), the imaginary part of coherence (imCoh), the corrected imaginary part of phase lagged value (ciPLV), the debiased weighted phase-locking index (dwPLI)) and three effective connectivity measures (generalized partial directed coherence (gPDC), direct directed transfer function (dDTF), and pairwise spectral granger prediction (pSGP)).

Methods:We compared the measures using non-dynamic and dynamic models of simulated EEG to assess their performance in the presence of three confounders: i) common input, ii) indirect connections, and iii) volume conduction. We also used experimental EEG to calculate i) the effect size of the difference between eyes-closed (EC) and eyes-open (EO) resting-state conditions and ii) the ratio of the inter-to-intra-subject variance in connectivity estimates.

Results:Among the functional connectivity measures, we observed that dwPLI is the least sensitive to volume conduction. dwPLI, along with ciPLV, better distinguished EC and EO in experimental EEG. dwPLI also displayed the highest inter-to-intra-subject variance ratio, meaning that dwPLI would more likely produce consistent measurements from one individual versus from different individuals. We observed that dDTF is the least sensitive to volume conduction. dDTF also best distinguished between EC and EO conditions, suggesting that it is better suited for analyses between two experimental conditions.

Discussion: Some of our results contradicted our hypotheses. Among the functional connectivity measures, Coh was less sensitive to common input compared to dwPLI. Among the effective connectivity measures, gPDC and pSGP outperformed dDTF in distinguishing direct from indirect connections. We discuss potential explanations for these results and ways to further validate them.

Conclusion: Our results allow us to make a few recommendations regarding connectivity measurement. We recommend using dwPLI as the primary measure for functional connectivity, followed up by estimates using ciPLV and Coh. dwPLI performed the best against volume conduction and distinguished between EC and EO conditions. Coh performed better than dwPLI and the other measures against common input and indirect connections. ciPLV distinguished better between EC and EO. We recommend using dDTF as the primary measure for effective connectivity, followed up by an estimate using gPDC. dDTF was the only measure that distinguished between EC and EO conditions. It performed better than pSGP against volume conduction and common input, but not significantly differently from gPDC.

Rights

© 2025, Diksha Srishyla

Download

Share

COinS